Automated adaptive computer support system generating problem solutions having confidence levels used to determine whether human expert intervention is required

Abstract

A solution engine of a vendor's highly-automated adaptive computer support system for a remote customer automatically generates proposed solutions, e.g., sets of support documents, as a function of diagnostic data received from a customer's computer system. The automatically generated solution can be subject to expert review prior to publication to the customer, e.g., when the automated system assigns a low confidence level to the solution. In addition, expert review can be triggered by feedback from the customer once a proposed solution is communicated. The diagnostic data, solutions and feedback for an incident are packaged as a “case” and entered into a historical case database. A solution function updater updates the solution function as a function, at least in part, of the expert review and customer feedback.

1. A computer support method comprising:
 applying plural solution functions in parallel to diagnostic data from a computer system being monitored; 
 calculating a confidence factor from the results of the plural solution functions; 
 if the confidence factor does not meet predefinable criteria, triggering an expert review of the results prior to delivery of the results to the computer system; and 
 delivering the results either directly or following modification by an expert to the computer system without direct contact between the expert and a computer system operator, 
 wherein the confidence factor indicates a likelihood that the results will be an effective solution for addressing a problem to which the diagnostic data is related. 
 
   
   
     2. A method as claimed in  claim 1  wherein one said plural solution functions employs a rule-based approach. 
   
   
     3. A method as claimed in  claim 1  wherein one of said plural solution functions employs a classification approach. 
   
   
     4. A method as claimed in  claim 1  wherein one of said plural solution functions employs a similarity-based approach. 
   
   
     5. A method as claimed in  claim 1 , wherein the predefinable criteria comprises a threshold value that the confidence factor is to meet before delivering the results without expert review. 
   
   
     6. A method as claimed in  claim 3  wherein said similarity-based approach accesses prototype cases. 
   
   
     7. A computer support system comprising:
 a solution engine for generating a result from diagnostic data by combining plural solutions generating using distinct artificial-intelligence approaches, said solution engine assigning a confidence factor to said result; 
 an expert review function for providing said result to an expert for review if said confidence level fails to meet predefinable criteria; 
 delivery means for delivering said result to a customer directly if said confidence factor meets said criteria and after modification by said expert if said confidence factor fails to meet said criteria without direct contact between the expert and the customer, 
 wherein the confidence factor indicates a likelihood that the results will be an effective solution for addressing a problem to which the diagnostic data is related. 
 
   
   
     8. A method as claimed in  claim 7  wherein said solution engine includes a rule-based engine. 
   
   
     9. A method as claimed in  claim 7  wherein said solution engine includes a classification engine. 
   
   
     10. A method as claimed in  claim 7  wherein said solution engine includes a similarity based engine. 
   
   
     11. A method as claimed in  claim 10  wherein said similarity-based engine also includes a database of prototype cases. 
   
   
     12. A computer support system comprising:
 a solution engine for generating a result from diagnostic data by combining plural solutions generating using distinct artificial-intelligence approaches, said solution engine assigning a confidence factor to said result; 
 an expert review function for providing said result to an expert for review if said confidence level fails to meet predefinable criteria; 
 delivery means for delivering said result to a customer directly if said confidence factor meets said criteria and after modification by said expert if said confidence factor fails to meet said criteria without direct contact between the expert and the customer, 
 wherein the predefinable criteria comprises a threshold value that the confidence factor is to meet before delivering the results without expert review. 
 
   
   
     13. A method as claimed in  claim 12  wherein said solution engine includes a rule-based engine. 
   
   
     14. A method as claimed in  claim 12  wherein said solution engine includes a classification engine. 
   
   
     15. A method as claimed in  claim 12  wherein said solution engine includes a similarity based engine. 
   
   
     16. A method as claimed in  claim 15  wherein said similarity-based engine also includes a database of prototype cases.